A New Multi-focus Image Fusion Method Using Principal

本科毕业设计论文题目多聚焦图像融合算法研究专业名称学生姓名指导教师毕业时间毕业 任务书一、题目多聚焦图像融合算法研究二、指导思想和目的要求本题目来源于科研，主要研究多聚焦图像的概念，学习多聚焦图像的常用融合算法，进而实现相关算法。

希望通过该毕业设计，学生能达到：1．利用已有的专业知识，培养学生解决实际工程问题的能力；2．锻炼学生的科研工作能力和培养学生团队合作及攻关能力。

三、主要技术指标1．学习多聚焦图像的特点；2．研究多聚焦图像的融合算法；3．实现多聚焦图像的融合。

四、进度和要求第01周----第02周： 参考翻译英文文献；第03周----第04周： 学习多聚焦图像的特点；第05周----第08周： 研究多聚焦图像的融合算法；第09周----第14周： 编写多聚焦图像的融合程序；第15周----第16周： 撰写毕业设计论文，论文答辩。

五、主要参考书及参考资料1.张德丰.MATLAB 数字图像处理[M].北京：机械工业出版社，2012.2. 敬忠良. 图像融合——理论与应用[M].北京：高等教育出版社，2010.3. 郭雷. 图像融合[M]. 北京：电子工业出版社，2011.4. 孙巍. 孙巍. 像素及多聚焦图像融合算法研究[D].长春：吉林大学，2008.5. 马先喜. 多聚焦图像融合算法研究[D].无锡：江南大学，2012.学生 指导教师 系主任 __ __设计论文摘要图像融合是将同一对象的两个或多个图像按一定规则合成为一幅图像。

其关键是抽取每幅源图像中的清晰区域，并将这些清晰区域以一定的规则融合起来，从而生成一幅清晰且信息量完整的融合图像。

多聚焦图像融合的具体目标在于提高图像的空间分辨率、改善图像的几何精度、增强特征显示能力、改善分类精度、替代或修补图像数据的缺陷等。

本文概括了多聚焦图像融合的一些基本概念和相关的基本知识，对DWT分解的层数和方向子带的个数对融合结果的影响进行了初步的研究。

1
概述
动态范围 （Dynamic Range， DR） 定义为图像中可表
多曝光图像融合使用不同长度曝光时间对同一场 景进行多次曝光采样， 分别获得不同强度范围入射光的 细节。通过将这些图像采样有效融合， 可以有效抑制随 机噪声， 获得高动态范围和低时域噪声图像。其主要目 标为正确反映出全局亮度差异， 保持全局对比度； 保留 亮、 暗区域尽可能多的原始细节； 对过饱和与欠曝光区 域颜色补偿修正。 Burt 等 [4]最早提出将图像融合技术用于动态范围扩 展。 Goshtasby 等 [5] 使用分块熵方法进行多曝光图像融 合， 块大小和块混合交接宽度与内容无关， 需要进行迭 代优化。 Mertens 等 [6] 在 RGB 空间采用对比度、 饱和度 和充分曝光度作为融合指导指标， 采用拉普拉斯金字塔
处理策略。在融合规则中提出全局动态范围系数来反映全局照度范围， 并指导亮度融合来突出融合结果的动态范 围； 使用局部对比度和色彩饱和度来反映原始图像的曝光程度， 并用于指导色彩融合； 使用小波分析作为多尺度融 合工具。对算法进行测试并与已有算法的结果进行信息熵和动态范围比较， 结果表明该算法对于信息熵和动态范 围均有提高。 关键词： 多次曝光图像融合； 动态范围； 小波分析 文献标志码： A 中图分类号： TP391.4 doi： 10.3778/j.issn.1002-8331.1203-0003
Computer Engineering and Applications 计算机工程与应用
2014， 50 （1）
153
大动态范围多曝光图像融合方法
胡燕翔， 万 莉
HU Yanxiang, WAN Li
天津师范大学 计算机与信息工程学院， 天津 300387 Computer and Information Engineering College, Tianjin Normal University, Tianjin 300387, China HU Yanxiang, WAN Li. Multi exposure image fusion based on dynamic range extending. Computer Engineering and Applications, 2014, 50 （1） ： 153-155. Abstract：A multi-scale multi exposure image fusion method based on dynamic range extending is proposed. Fusion goal, composed strategies of global dynamic range, local contrast and color are discussed in detail. Global dynamic range factor is used to measure the global illumination range and guide the intensity fusion； local contrast and color saturation are extracted and used to guide color fusion. Wavelet is used as multi-scale analysis tool. The algorithm is tested using Matlab and compared with existing algorithms. The results show that the algorithm can get wide dynamic range while keeping high entropy. Key words： multi exposure fusion; dynamic range; wavelet analysis 摘 要： 提出一种基于动态范围扩展的多曝光图像多尺度融合方法。讨论融合目标与动态范围分布、 细节及颜色的

Abstract: A new micro 3D reconstruction method based on multi-focus image fusion model was proposed which overcame the disadvantage of low efficiency in the processing of the laser scanning confocal microscopy metal samples. In the method, the micro 3D reconstruction problem was transformed into a two-dimensional image fusion problem. The whole process can be divided into three stages, including image fusion, optimal matching and iterative recovery stages. In the first phase, a novel microscopy image fusion algorithm based on m-PCNN in non-subsampled Contourlet transform (NSCT) domain was proposed, in which details of each original image can be retained to the first fusion result. In the next phase, the second fusion result and the height map were obtained by an optimal matching algorithm based on the correlation coefficient of regional image. In the last phase, an en收稿日期 : 2016-09-19; 修回日期 : 2017-02-14. 基金项目 : 国家 “ 八六三 ”高技术研究发展计划 (2015AA015408); 中国科学院西部 之光基金 (2011180). 闫 涛 (1987 — ), 男 , 博士研究生 , 主要研究方向为图像处理与机器视觉、智能计算 ; 陈 斌 (1970— ), 男 , 博士 , 研究员 , 博士生导师 , 论文通讯作者 , 主要研究方向为图像理解、三维视觉分析 ; 刘凤娴 (1987 — ), 女 , 博士研究生 , 主要研究方向为 大气环境化学 ; 王佐才 (1987 — ), 男 , 博士研究生 , 主要研究方向为图像处理 ; 郭玉峰 (1985 — ), 男 , 博士研究生 , 主要研究方向为机 器视觉方法 ; 郭四稳 (1965— ), 男 , 博士 , 教授 , 博士生导师 , 主要研究方向为图像处理与计算机自动推理、数字图像处理 .

2021574彩色图像灰度化是图像处理和计算机视觉领域的基本课题和重要前提，是将三维通道信息转换为一维灰度数据的过程。

为了节约成本，人们仍使用黑白打印，并且许多出版物的大部分图片是灰度图像。

生活中还存在很多更有艺术效果的黑白图像，由此衍生了灰度图像在艺术美学方面的应用，如中国水墨画渲染、黑白摄影等[1]。

为了减少输入图像的信息量或者减少后续的运算量，都需要将彩色图像进行灰度化处理，其在图像预处理等方面有很多应用，如边缘检测[2-3]、特征提取[4-5]等。

为了使灰度化后的图像更好地保留彩色图像特征，许多方法被相继提出。

根据算法中映射函数是否可应用于整幅图像的所有像素，常见的灰度化算法大致可以分为两类：全局映射法和局部映射法。

在局部映射法中，灰度值随着空间位置而改变，将不同的灰度值赋给相同的颜色以增强灰度图像局部对比度，容易受相邻像素的影响。

2004年，Bala等人[6]将高频色度信息引入亮度通道，局部保留了相邻颜色之间的差异。

Smith等人[7]使用拉普拉斯金字塔提取图像多层特征，根据彩色与灰度图像色对比度比例来调整拉普拉斯各分层灰度值，增强不明显的边缘，进行对比度调整。

卢红阳等人[8]提出一种基于最大加权投影求解的算法，建立最大化的加权局部保留投影模型，提出最大加权投影的目标优化函数。

局部映射法试图找出颜色在三维的局部差异，通过控制像素的亮度值，从而精确地保留图像的局部特征，但无法保证全局颜色的一致性，把一个整体图像转换为非齐次的，最终求出的灰度图像彩色图像多尺度融合灰度化算法顾梅花，王苗苗，李立瑶，冯婧西安工程大学电子信息学院，西安710600摘要：为了使彩色图像灰度化后能够保留更多的原始特征，提出了一种新的基于多尺度图像融合的灰度化算法。

将彩色图像分解为R、G、B三个通道图像，采用基于高斯-拉普拉斯金字塔的多尺度图像融合模型进行灰度化，并引入梯度域导向图像滤波（Gradient Domain Guided Image Filter，GGIF）来消除多尺度融合可能产生的伪影。

一种新的结合SVM和FNN的多聚焦图像融合算法：To deal with the problems of cracks among blocks and the uncertainty of real characteristics of image fusion based on block， this paper proposed a new multifocus image fusion method by combining support vector machine （SVM）wits fuzzy neural network （FNN）. Firstly， FCM and SVM were used to obtain the parameters of FNN and the block was divided into clear， blurring and transitional zones based on the FNN. Then the three classified areas were merged with weighting to get the fused multifocus images， where the weight factors were obtained as the defuzzication outputs of the fuzzy neural network. Finally， the qualities of various fusion algorithm were evaluated by the root mean square error（RMSE）， the mean absolute error（MAE） and peak signal to noise ratio（PSNR）. The experimental results show that the proposed fusion algorithm has good robustness and computing performance， which basically meets the demand of practical image fusion， and the fusion quality evaluations illustrate that our method has an advantage over the existing fusion algorithm.1 引言.图像信息融合是信息融合的一个重要分支，通过对同一场景下多个传感器获取的图像信息进行有机的综合，生成信息量更加全面、精确、完整的图像，以弥补单一传感器获取信息的局限性[1]。

基于GHM多小波变换的非织造布多焦面图像融合陈阳;辛斌杰;邓娜【摘要】针对光学显微镜在单一焦平面下拍摄的织物图像部分区域纤维会模糊的问题,提出基于GHM多小波变换的非织造布多焦面图像融合算法.利用自行搭建的非织造布显微成像系统采集不同焦平面下的织物图像序列,对初始图像序列进行临界采样预滤波处理,使用2种融合方法逐一处理图像的高低频,初始织物融合图像经多小波融合及逆变换后获得,之后按上述方法将初始融合图像与后续单焦面图像融合,叠加循环至融合后所有纤维区域均能清晰显示为止结束收敛.实验结果表明,该融合方法能将不同焦平面下拍摄的图像序列进行数字化图像融合,达到单幅图像内全视野区域的纤维网清晰聚焦融合的效果,为之后的计算机图像处理及测量提供便利.【期刊名称】《纺织学报》【年(卷),期】2019(040)006【总页数】8页(P125-132)【关键词】临界采样预滤波;GHM多小波;多焦面融合;非织造布图像;显微成像【作者】陈阳;辛斌杰;邓娜【作者单位】上海工程技术大学电子电气工程学院,上海 201620;上海工程技术大学服装学院,上海 201620;上海工程技术大学电子电气工程学院,上海 201620【正文语种】中文【中图分类】TP311.1非织造布是由纤维随机或定向排列而成，生产以纺黏法为主。

非织造布的性能与纤维网的孔隙构造紧密相关，而纤维的厚度、宽度、取向度以及纤维网的形成方式等都与其构造有关，因此能够得到这些结构参数进而找到性能间的联系，对生产及用途都具有十分重要的指导意义。

目前主要使用间接法对非织造布孔隙结构进行解析，而其存在的问题集中在费时费力且不能考虑到孔结构的复杂性。

计算机数字图像处理技术的发展为研究非织造布结构和性能提供了有效工具。

图像质量对纤维的形态测量与结构解析至关重要。

非织造布的厚度太大使得一般光学显微镜的景深不足以将所有纤维清晰地显现在一幅图像中。

基于这种不完全聚焦图像的测量，纤维结构将是不准确的，甚至会对后续处理有一定的误导性[1]。

基于Ｃｏｎｔｏｕｒｌｅｔ变换的多聚焦图像融合作者：丁岚来源：《电脑知识与技术》2008年第34期摘要：由于可见光成像系统的聚焦范围有限，很难获得同一场景内所有物体都清晰的图像，多聚焦图像融合技术可有效地解决这一问题。

Contourlet变换具有多尺度多方向性，将其引入图像融合，能够更好地提取原始图像的特征，为融合图像提供更多的信息。

该文提出了一种基于区域统计融合规则的Contourlet变换多聚焦图像融合方法。

先对不同聚焦图像分别进行Contourlet变换，采用低频系数取平均，高频系数根据区域统计值决定的融合规则，再进行反变换得到融合结果。

文中给出了实验结果，并对融合结果进行了分析比较，实验结果表明，该方法能够取得比基于小波变换融合方法更好的融合效果。

关键词：图像融合；Contourlet变换；小波变换中图分类号：TP391文献标识码：A文章编号：1009-3044(2008)34-1700-03Multifocus Image Fusion Based on Contorlet TransformDING Lan(College of Information Science & Technology, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China)Abstract: Due to the limited depth-of-focus of optical lenses , it is often difficult to get an image that contains all relevant objects in focus. Multifocus image fusion method can solve this problem effectively. Contoulet transform has varying directions and multiple scales. When the contourlet transform is introduced to image fusion , the characteristics of original images are taken better and more information for fusion is obtained. A new multifocus image fusion method is proposed in this paper, based on contourlet transform with the fusion rule of region statistics. Different focus images are decomposed using contourlet transform firstly, then low-bands are integrated using the weighted average , high-bands are integrated using region statistics rule. Then the fused image will be obtained by inverse contourlet transform. The experimental results are showed, and compared with the method based on wavelet transform. Experiments show that this approach can achieve better results than the method based on wavelet transform.Key words: image fusion; contourlet transform; wavelet transform1 引言对于可见光成像系统来讲，由于成像系统的聚焦范围有限，场景中的所有目标很难同时都成像清晰，这一问题可以采用多聚焦图像融合技术来解决，即用同一成像镜头对场景中的两个（多个）目标分两次（多次）进行成像，将这些成像中清晰部分融合成一幅新的图像，以便于人的观察或计算机的后续处理。

J. Wang, X. Liao, and Z. Yi (Eds.): ISNN 2005, LNCS 3497, pp. 753–758, 2005.© Springer-Verlag Berlin Heidelberg 2005Multifocus Image Fusion Using Spatial Featuresand Support Vector MachineShutao Li 1,2 and Yaonan Wang 11 College of Electrical and Information Engineering, Hunan UniversityChangsha, Hunan 410082, China 2 National Laboratory on Machine Perception, Peking University, Beijing 100871, China *******************.cnAbstract. This paper describes an application of support vector machine to pixel-level multifocus image fusion problem based on the use of spatial features of image blocks. The algorithm first decomposes the source images into blocks. Given two of these blocks (one from each source image), a SVM is trained to determine which one is clearer. Fusion then proceeds by selecting the clearer block in constructing the final image. Experimental results show that the pro-posed method outperforms the discrete wavelet transform based approach, par-ticularly when there is movement in the objects or misegistration of the source images.1 IntroductionOptical lenses often suffer from the problem of limited depth of field. Consequently, the image obtained will not be in focus everywhere. A possible way to alleviate this problem is by image fusion [1], in which several pictures with different focus parts are combined to form a single image. This fused image will then hopefully contain all relevant objects in focus.In recent years, various methods based on multiscale transforms have been pro-posed, including the Laplacian pyramid [2], the gradient pyramid [1], the ratio of Low pass pyramid [3] and the morphological pyramid [4]. More recently, the discrete wavelet transform (DWT) [5], [6] has also been used. In general, DWT is superior to the previous pyramid based methods [6]. While these methods often perform satisfac-torily, their multiresolution decompositions and consequently the fusion results are not shift invariant because of an underlying down sampling process. When there is slight camera/object movement or when there is misregistration of the source images, their performance will thus quickly deteriorate.In this paper, we propose a pixel level multifocus image fusion method based on the use of spatial features of image blocks and support vector machines (SVM). The implementation is computationally simple and is robust to shift problem. Experimen-tal results show that it outperforms the DWT based method. The rest of this paper is organized as follows. The proposed fusion scheme will be described in Section 2. Experiments will be presented in Section 3, and the last section gives some conclud-ing remarks.754 Shutao Li and Yaonan Wang2 SVM Based Multifocus Image Fusion2.1 Feature ExtractionIn this paper, we extract two measures from each image block to represent its clarity. These are described in detail as follows.2.1.1 Spatial Frequency (SF)Spatial frequency is used to measure the overall activity level of an image [7]. For an N M × image F , with the gray value at pixel position (m ,n ) denoted by F (m ,n ), its spatial frequency is defined as22CF RF SF +=. (1)where RF and CF are the row frequency∑∑==−−=M m N n n m F n m F MN RF 122))1,(),((1, and column frequency∑∑==−−=N n M m n m F n m F MN CF 122)),1(),((1,respectively. 2.1.2 Absolute Central Moment (ACM) [8])(10i p u i ACM I i ∑−=−=.(2)where µis the mean intensity value of the image, and i is the gray level.2.1.3 Demonstration of the Effectiveness of the MeasuresIn this section, we experimentally demonstrate the effectiveness of the two focus features. An image block of size 64h 64 (Fig. 2(a)) is extracted from the “Lena” im-age. Fig. 2(b) to Fig. 2(e) show the degraded versions by blurring with a Gaussian filter of radius 0.5, 0.8, 1.0 and 1.5 respectively. As can be seen from Table 1, when the image becomes more blurred, the two features are monotonic accordingly. These results suggest that both two features can be used to reflect image clarity.2.2 The Fusion AlgorithmFig.2 shows a schematic diagram of the proposed multifocus image fusion method. Here, we consider the processing of just two source images, though the algorithm can be extended straightforwardly to handle more than two.Multifocus Image Fusion Using Spatial Features and Support Vector Machine 755(a) original region (b) radius=0.5 (c) radius=0.8 (d) radius=1.0 (e) radius=1.5Fig. 1. Original and blurred regions of an image block extracted from “Lena”Table 1. Feature values for the image regions in Fig.1.Fig. 1(a) Fig. 1(b) Fig. 1(c) Fig. 1(d) Fig. 1(e) SF 40.88 20.61 16.65 14.54 11.70 ACM 51.86 48.17 47.10 46.35 44.96Fig. 2. Schematic diagram of the proposed fusion method.In detail, the algorithm consists of the following steps:1. Decompose the two source images A and B into blocks with size of M×N. De-note the i th image block pair by Aiand Birespectively.2. From each image block, extract two features above described that reflect its clar-ity. Denote the feature vectors for Aiand Biby (iiAAACMSF,) and (iiBBACMSF,)re-spectively.3. Train a SVM to determine whether Aior Biis clearer. The difference vector),(iiiiBABAACMACMSFSF−− is used as input, and the output is labeled accordingto=otherwiseanclearer thisif1target iiiBA. (3)4. Perform testing of the trained SVM on all image block pairs obtained in Step 1.The i th block, Zi, of the fused image is then constructed as>=otherwise5.0ifiiii BoutAZ. (4)where outiis the SVM output using the i th image block pair as input.756 Shutao Li and Yaonan Wang5. Verify the fusion result obtained in Step 4. Specifically, if the SVM decides thata particular block is to come from A but with the majority of its surrounding blocks from B, this block will be switched to come from B. In the implementation, a majority filter with a 3×3 window is used.3 ExperimentsExperiment is performed on 256 level source images shown in Fig.3(a) and (b). Their sizes are 512h512. The true gray value of the reference image is not available and so only a subjective visual comparison is intended here. Image blocks of size 32h32 are used. Two pairs of regions in Fig.3(a),(b), each containing 18 image block pairs, are selected as training set. In 9 of these block pairs, the first image is clearer than the second image, and the reverse is true for the remaining 9 pairs. The two spatial fea-tures are extracted and normalized to the range [0,1] before feeding into SVM. In the experiment, linear kernel is used.For comparison purposes, we also perform fusion using the DWT. The wavelet ba-sis “db8”, together with a decomposition level of 5, is used. Similar to [6], we employ a region based activity measurement for the active level of the decomposed wavelet coefficients, a maximum selection rule for coefficient combination, together with a window based consistency verification scheme.(a) focus on the ‘Pepsi’ can. (b) focus on the testing card.(c) fused image using DWT (db8, level=5). (d) fused image using SVM.Fig. 3. The “Pepsi” source images and fusion results. The training set is selected from regions marked by the rectangles in Fig. 3(a) and Fig. 3(b).Multifocus Image Fusion Using Spatial Features and Support Vector Machine 757(a)difference between the fused image usingDWT (Fig.3(c)) and source image Fig.3(a). (b)difference between the fused image using DWT (Fig.3(c)) and source image Fig.3(b).(c) difference between the fused image usingSVM (Fig. 3(d)) and source image Fig.3(a). (d)difference between the fused image using SVM (Fig.3(d)) and source image Fig. 3(b). Fig. 4. Differences between the fused images in Fig.3(c),(d) and source images in Fig.3(a),(b). Fusion results on using DWT and SVM are shown in Fig. 3(c),(d). Take the “Pepsi” images as an example. Recall that the focus in Fig. 3(a) is on the Pepsi can while that in Fig. 3(b) is on the testing card. It can be seen from Fig.3(d) that the fused image produced by SVM is basically a combination of the good focus can and the good focus board. In comparison, the result by DWT shown in Fig.3(c) is much infe-rior. Clearer comparisons of their performance can be made by examining the differ-ences between the fused images and each source image (Fig.4).4 ConclusionsIn this paper, we proposed a method for pixel level multifocus image fusion by using the spatial features of image blocks and SVM. Features indicating the clarity of an image block, namely, spatial frequency and absolute central moment, are extracted and fed into the support vector machine, which then learns to determine which source image is clearer at that particular physical location. Experimental results show that this method outperforms the DWT based approach, particularly when there is object movement or registration problems in the source images.758 Shutao Li and Yaonan WangAcknowledgementsThis work is supported by the National Natural Science Foundation of China (No.60402024).References1. Burt, P.J., Kolczynski, R.J.: Enhanced Image Capture through Fusion. In Proc. of the 4thInter. Conf. on Computer Vision, Berlin (1993) 173-1822. Burt, P.J., Andelson, E.H.: The Laplacian Pyramid as a Compact Image Code. IEEE Trans.Comm., 31 (1983) 532-5403. Toet, A., Ruyven, L.J., Valeton, J.M.: Merging Thermal and Visual Images by a ContrastPyramid. Optic. Eng., 28 (1989) 789-7924. Matsopoulos, G.K., Marshall, S., Brunt, J.N.H.: Multiresolution Morphological Fusion ofMR and CT Images of the Human Brain. Proc. of IEE: Vision, Image and Signal, 141 (1994) 137-1425. Li, H., Manjunath, B.S., Mitra, S.K.: Multisensor Image Fusion using the Wavelet Trans-form. Graph. Models Image Proc., 57 (1995) 235-2456. Zhang, Z., Blum, R.S.: A Categorization of Multiscale-Decomposition-Based Image FusionSchemes with a Performance Study for a Digital Camera Application. Proc. of the IEEE, 87 (1999) 1315-13257. Eskicioglu, A.M., Fisher, P.S.: Image Quality Measures and Their Performance. IEEE Trans.Comm., 43 (1995) 2959-29658. Shirvaikar, M.V.: An Optimal Measure for Camera Focus and Exposure. 36th IEEE South-eastern Symp. on Sys. Theory, Atlanta (2004) 472-475。

基于可控滤波器和空间频率的图像融合算法郭峰;杨静;史健芳【摘要】A multi-focus image fusion algorithm based on steerable filters and spatial frequency was proposed by researching of the pixel-level image fusion.Steerable filters were used to filter the original image to obtain the oriented analytic image in diffe-rent directions.To obtain the oriented analytic spatial frequency map,the local spatial frequency of each oriented analytic image was calculated and compared.The fusion result image was generated with pixels which were selected form original images ac-cording to the fusion rule.Experimental results show that,compared with the results of other fusion algorithms,the proposed algorithm has certain advantages in the subj ective evaluation and obj ective evaluation indicators.%对像素级图像融合进行研究，提出一种基于可控滤波器和空间频率的多聚焦图像融合算法。

利用可控滤波器处理源图像，得到源图像在不同方向上的解析图像，计算并比较方向解析图像的局部空间频率，得到源图像的方向解析空间频率谱，通过融合规则选取源图像中的像素构成融合图像。

基于分块的不可分小波多聚焦图像融合刘维杰1，刘 斌2，彭嘉雄3(1. 武汉大学计算机学院，武汉 430079；2. 湖北大学数学与计算机科学学院，武汉 430062；3. 华中科技大学图像识别与人工智能研究所，武汉 430074)摘 要：为解决现有图像融合方法存在均方根误差较大、熵值和空间频率较小的问题，提出一种基于分块的不可分小波多聚焦图像融合方法。

该方法利用不可分小波滤波器组对原图像进行多尺度分解，选取特征较明显(方差大)的块作为融合子图像的组成块，对融合块图像做不可分小波逆变换后形成融合图像。

实验结果表明，相比其他融合方法，该方法能消除块痕迹、节约运算量，具有更好的融合效果。

关键词：多聚焦图像融合；不可分小波；均方根误差Multi-focus Image Fusion of Nonseparable WaveletBased on BlockingLIU Wei-jie 1, LIU Bin 2, PENG Jia-xiong 3(1. School of Computer, Wuhan University, Wuhan 430079, China; 2. School of Mathematics and Computer Science,Hubei University, Wuhan 430062, China; 3. Institute of Image Recognition and Artificial intelligence,Huazhong University of Science and Technology, Wuhan 430074, China)【Abstract 】In order to solve the problems that the existing image fusion methods have bigger Root Mean Square Error(RMSE) and smaller entropy and spatial frequency, this paper presents a multi-focus image fusion of nonseparable wavelet based on blocking. The sources images are decomposed using the nonseparable wavelet the filter bank. Then the subimages are segmented into blocks, and these blocks are fused by selecting the bigger value of the variance of the block. The inverse wavelet transform is carried out to produce the fused image. Experimental result shows that the fusion performance of the method is better than the other fusion methods. It can eliminate the blocking artifacts of the fused images and save the time of fusion.【Key words 】multi-focus image fusion; nonseparable wavelet; Root Mean Square Error(RMSE) DOI: 10.3969/j.issn.1000-3428.2011.02.071计 算 机 工 程 Computer Engineering 第37卷 第2期V ol.37 No.2 2011年1月January 2011·图形图像处理· 文章编号：1000—3428(2011)02—0205—02文献标识码：A中图分类号：TP3911 概述近年来，图像融合在军事领域和非军事领域如遥感图像、医学图像、机器视觉上得到了广泛应用[1]。

医疗影像技术中的新技术与新方法近年来，医学与技术的融合日趋紧密，医疗影像技术也正日新月异。

随着新技术的不断涌现和更新，医疗影像技术的应用范围也越来越广泛，为医学诊断与治疗带来了无限的可能性。

本文将从各个角度出发，探讨医疗影像技术中的新技术与新方法，以期读者对此有所收获。

一、计算机辅助诊断技术（Computer-Aided Diagnosis, CAD）计算机辅助诊断技术是一种以计算机为核心，通过数字图像处理、特征提取、分类诊断等方法来辅助医师完成疾病的诊断和鉴别诊断的新技术。

随着计算机技术的不断发展，CAD在肺部结节、乳腺癌、肝癌、脑部疾病、心血管疾病等多个领域显示了一定的优势。

例如，基于CAD技术的乳腺三维造影，可以获得大量高分辨率的乳腺图像，并运用计算机算法来寻找信号异常，为人工判读提供一系列的处理手段，从而提高乳腺癌早期诊断的准确率和可信度。

二、医疗影像加强技术（Medical Image Enhancement）医疗影像加强技术是指运用各种图像处理算法对原始影像进行处理以提高影像质量的一系列技术。

该技术可以通过消除影像中的噪声、增强图像对比度、改善图像分辨率等手段来提高图像质量，并在医学诊断、手术规划等方面发挥重要作用。

目前，医疗影像加强技术在核磁共振影像（MRI）、计算机断层扫描（CT）、放射性核素影像等多个医疗领域得到了广泛应用。

例如，针对肺部 CT 影像，借助医疗影像加强技术，可以获得更加清晰的肺组织影像，有利于医生对肺部疾病进行更加精准的诊断。

三、多模态医疗影像融合技术（Multi-Modal Medical Image Fusion）多模态医疗影像融合技术，是指将不同成像模态得到的医疗影像进行融合，从而得到更全面、准确的医疗影像信息的一种新技术。

通过对不同成像技术的综合应用，该技术可以获得更多的医疗影像信息，将不同模态的影像融合在一起，获得更加完整、准确、可靠的影像信息，提高医疗诊断的准确性和可信度。

shearlet变换和区域特性相结合的图像融合郑伟;孙雪青;李哲【摘要】为了提高多模医学图像或多聚焦图像的融合性能，结合shearlet变换能够捕捉图像细节信息的性质，提出了一种基于shearlet变换的图像融合算法。

首先，用shearlet变换将已精确配准的两幅原始图像分解，得到低频子带系数和不同尺度不同方向的高频子带系数。

低频子带系数使用改进的加权融合算法，用平均梯度来计算加权参量，以此来改善融合图像轮廓模糊度高的问题，高频子带系数采用区域方差和区域能量相结合的融合规则，以得到丰富的细节信息。

最后，进行shearlet逆变换得到融合图像。

结果表明，此算法在主观视觉效果和客观评价指标上优于其它融合算法。

%In order to improve the performance of multi-modality medical image fusion and multi-focus image fusion, since the shearlet transform can capture the detail information of images, an image fusion algorithm based on shearlet transform was proposed.Firstly, the shearlet transform was used to decompose the two registered original images, thus the low frequency sub-band coefficients and high frequency sub-band coefficients of different scales and directions were obtained.The fusion principle of low frequency sub-band coefficients was based on the method of weighted fusion, using the average gradient to calculate the weighted parameters in order to improve the edge fuzzy of the fused image.As for the high frequency sub-band coefficients, a fusion rule adopting the region variance combining with the region energy to get the detail information was presented.Finally, the fused image was reconstructed by inverse shearlet transform.The results show that thealgorithm is superior to other fusion algorithms on subjective visual effect and objective evaluation.【期刊名称】《激光技术》【年(卷),期】2015(000)001【总页数】7页(P50-56)【关键词】图像处理;图像融合;shearlet变换;加权融合;区域方差;区域能量【作者】郑伟;孙雪青;李哲【作者单位】河北大学电子信息工程学院，保定071002; 河北大学河北省数字医疗工程重点实验室，保定071002;河北大学电子信息工程学院，保定071002; 河北大学河北省数字医疗工程重点实验室，保定071002;河北大学电子信息工程学院，保定071002; 河北大学河北省数字医疗工程重点实验室，保定071002【正文语种】中文【中图分类】TP391E-mail:*********************图像融合是将两幅或两幅以上的图像合成一幅新图像的过程，但这并不意味着仅仅是把多幅图像相加，而是要使用特定的规则把每幅图像的有用信息综合起来，以获取一幅更加准确、更加全面的新图像。

◦ IHS颜色空间，即亮度(I)、色度(H)、饱和度(S)，
三分量之间互不相关，因此对某一分量的处理不会
影响到其它分量。而三分量中，以亮度分量最为重
要。R
I
G
IHS 变换
H
H、S反差 扩展
I’
R’
H’
IHS 反变换
G’
彩 色 合
B
S
I分量
S’
B’
成
Landsat-TM
SPOT Pan
精品课件
8
PCA(主分量分析)变换融合法
Model-Based Satellite Image Fusion. Multi-band SAR Images Fusion Using the EM Algorithm
in Contourlet Domain. Multifocus Holographic 3-D Image Fusion Using
Preserve Spatial Details and Minimize Color Distortion. Context-driven fusion of high spatial and spectral
resolution images based on oversampled multiresolution analysis. Fusion of multispectral and panchromatic images using improved IHS and PCA mergers based on wavelet decomposition. Fusion of Multispectral and Panchromatic Images Using
23

I.J. Image, Graphics and Signal Processing, 2013, 12, 56-61Published Online October 2013 in MECS (/)DOI: 10.5815/ijigsp.2013.12.08Correcting Multi-focus Images via SimpleStandard Deviation for Image FusionFiras A. JassimManagement Information Systems Department, Irbid National University, Irbid 2600, Jordanfirasajil@Abstract— Image fusion is one of the recent trends in image registration which is an essential field of image processing. The basic principle of this paper is to fuse multi-focus images using simple statistical standard deviation. Firstly, the simple standard deviation for the k×k window inside each of the multi-focus images was computed. The contribution in this paper came from the idea that the focused part inside an image had high details rather than the unfocused part. Hence, the dispersion between pixels inside the focused part is higher than the dispersion inside the unfocused part. Secondly, a simple comparison between the standard deviation for each k×k window in the multi-focus images could be computed. The highest standard deviation between all the computed standard deviations for the multi-focus images could be treated as the optimal that is to be placed in the fused image. The experimental visual results show that the proposed method produces very satisfactory results in spite of its simplicity.Index Terms— Image fusion, Multi-focus, Multi-sensor, Standard deviation.I.INTRODUCTIONImage fusion is the process of integrating two or more images to construct one fused image which is highly informative. In accordance with the development of technology and inventing various image-capturing devices, it is not always possible to get a detailed image that contains all the required information. During the past few years, image fusion gets greater importance for many applications especially in remote sensing, computer vision, medical imaging, military applications, and microscopic imaging [27]. The multi-sensor data in the field of remote sensing, medical imaging and machine vision may have multiple images of the same scene providing different information. In machine vision, due to the limited depth-of-focus of optical lenses in charge-coupled device (CCD), it is not possible to have a single image that contains all the information of objects in the image. Sometimes, a complete picture may not be always feasible since optical lenses of imaging sensor especially with long focal lengths, only have a limited depth of field [9]. The basic objective image fusion is to extract the required information of an image that was captured from various sources and sensors. After that, fuse these images into single fused (composite) enhanced image [14]. In fact, the images that are captured by usual camera sensors have uneven characteristics and outfit different information. Therefore, the essential usefulness of image fusion is its important ability to realize features that can not be realized with traditional type of sensors which leads to enhancing human visual perception.In fact, the fused image contains greater data content for the scene than any one of the individual image sources alone [7]. Image fusion helps to get an image with all the information. The fusion process must preserve all relevant information in the fused image.This paper discusses the fusion of multi-focus images using the simple statistical standard deviation in k×k window size. Section II gives a review of the recent work and produces variety of different approaches reached. The proposed technique for fusing images had been presented in section III. The experimental results and performance assessments are discussed in section IV. At the end, the main conclusions of this paper have been discussed in section V.II.IMA GE FUSION PRELIMINA RIESThe essential seed of image fusion goes back to the fifties and sixties of the last century. It was the first time to search for practical methods of compositing images from different sensors to construct a composite image which could be used to better coincide natural images. Image fusion filed consists of many terms such as merging, combination, integration, etc [26]. Nowadays, there are two approaches for image fusion, namely spatial fusion and transform fusion. Currently, pixel based methods are the most used technique of image fusion where a composite image has to be synthesized from several input images [1][21]. A new multi-focus image fusion algorithm, which is on the basis of the ratio of blurred and original image intensities, was proposed by [19]. A generic categorization is to consider a process at signal, pixel, or feature and symbolic levels [15]. An application of artificial neural networks to solve multi-focus image fusion problem was discussed by [14]. Moreover, the implementation of graph cuts into image fusion to support was proposed by [16]. According to [23], a compressive sensing image fusion had been discussed.A wavelet-based image fusion tutorial must not beforgotten and could be found in [18]. A complex wavelets and its application to image fusion can be found in [11]. Image fusion using principle components analysis (PCA) in Compressive sampling domain could be found [28]. A new Intensity-Hue-Saturation (HIS) technique to Image Fusion with a Tradeoff Parameter was discussed by [3]. An implementation of Ripplet transform into medical images was discussed by [6]. An excellent comparative analysis of image fusion methods can be found in [26]. A comparison of different image fusion techniques for individual tree crown identification using quickbird images [20].Here, in this paper, in order to compare the proposed technique with some common techniques in image fusion, therefore; we have used the wavelet and principle components analysi s (PCA) as the most two widely implemented methods in image fusion [28]. The wavelet transform has become a very useful tool for image fusion. It has been found that wavelet-based fusion techniques outperform the standard fusion techniques in spatial and spectral quality, especially in minimizing color distortion. Schemes that combine the standard methods (HIS or PCA) with wavelet transforms produce superior results than either standard methods or simple wavelet-based methods alone. However, the tradeoff is higher complexity and cost [18]. On the other hand, principal component analysis is a statistical analysis for dimension reduction. It basically projects data from its original space to its eigenspace to increase the variance and reduce the covariance by retaining the components corresponding to the largest eigenvalues and discarding other components. PCA helps to reduce redundant information and highlight the components with biggest influence [26]. Furthermore, many tinny and big modifications on the known wavelet method were researched and discussed by many authors. On of these contributions the one that was introduced for image fusion named wavelet packet by [2]. Another contribution on wavelet called Dual Tree Complex Wavelet Transform (DT-CWT) can be found in [10][12]. Also, a curvelet image fusion was presented by [4][17]. Furthermore, many other fusion methods like contourlet by [8][22], and Non-subsampled Contourlet Transform (NSCT) which can be found in [5].III.PORPOSED TECHNIQUEThe proposed technique in this paper was based on the implementation of the simple standard deviation for each of the fused and input images. At the beginning, a brief description about statistical standard deviation must be introduced. Actually, the standard deviation is a measure of how spreads out numbers are. Moreover, it is the measure of the dispersion of a set of data from its mean. The more spread apart the data, the higher the deviation. The standard deviation has proven to be an extremely useful measure of spread in part because it is mathematically tractable. In this paper, the essential contribution that was used as a novel image fusion technique was based on the fact that the standard deviation in the focused part inside an image had high details rather than its identical part in the unfocused image. Hence, the dispersion between pixels inside the focused part is higher than the dispersion inside the unfocused part. Hence, the standard deviation could be computed for each k×k window inside all the input multi-focus images. The higher value of the standard deviation between all the computed standard deviations in all the input images may be treated as the optimal standard deviation. According to Fig. 1, the standard deviation for an arbitrary 2×2 window size was computed to show that the higher standard deviation value in the four input images is the image that has more details. Statistically speaking, the computed standard deviations for the four input images were 0.5, 1.7078, 3.304, and 2.63, respectively; for figures (1.a), (1.b), (1.c), and (1.d). Therefore, the higher value of the standard deviation is (3.304) which belong to the third image (Fig. 1.c) which has sharp edges rather than the other images with blurred or semi-blurred edges. Consequently, the k×k window with the high standard deviation value is recommended to be the optimal window and that is to be placed in the fused image from all the input images. However, the suitable value of the window size k is preferred to be as small as possible. Therefore, in this article, the best recommended k that is to be gives best results is 2.124 123132 130Figure 1. Four multi-focus images (a), (b), (c) and (d), eachwith its own standard deviationFigure 2. Blurred Block from Clock (Blur)188 188 186 185 186 184 182 180 179 178 188 187 186 185 184 183 182 182 181 179 187 188 187 186 184 183 182 182 181 179 188 188 187 187 186 184 182 180 179 178 188 187 187 186 185 184 181 179 176 175 187 186 185 184 182 182 180 178 174 172 186 185 183 182 181 180 178 176 173 171 186 185 183 181 181 180 177 174 172 170 184 183 182 181 179 177 174 173 171 168 180 180 179 178 177 174 171 170 167 164 Figure 3. Original Block from Clock (Sharp)189 189 188 186 185 183 181 182 183 182 190 190 190 189 186 183 181 182 182 181 188 188 188 188 187 184 182 181 181 181 188 187 187 187 188 186 183 182 182 182 189 188 189 189 187 186 184 182 182 182 188 187 187 188 186 185 184 182 181 181 187 186 185 185 185 185 184 183 182 181 188 187 187 187 185 186 186 184 182 181 186 187 188 188 187 183 182 183 182 181 184 184 185 186 185 181 182 182 179 176 Figure 4. Blurred Block from Clock (Blur)As another example, for the matter of changing the window size, a 10×10 window size was used from clock multi-focus image, Fig. 2. The computed standard deviations for the 10×10 window size were (3.2489) and (1.8803), for the sharp and blurred images (for the small clock), respectively. The 10×10 window size was indicated by wide dark borders. Once again it is clear that the sharp edges image has higher standard deviation than the blurred image. The 10×10 windows sizes for the sharp and the blurred images were presented in figures (3) and (4), respectively.IV.EXPERIMENTA L RESULTSIn this section, experimental results that support the proposed technique in this paper were presented and discussed. Several standard test images were used as input images and the resulted fused images were obtained according to the proposed technique by implementing the application of the highest standard deviation value. Obviously, the ocular results show that the proposed technique produces quite satisfactory results compared with other methods like wavelet transform and principle components analysi s (PCA). Practically, there are many performance measures used in image fusion techniques like Root Mean Square Error (RMSE), Peak Signal-to-Noise ratio (PSNR), Image Quality Index (IQI) [24], and structure similarity index (SSIM) [25][29]. Here, in this paper, the SSIM measure was implemented because of its high confidentiality [13]. TABLE1COMPARISON OF SSIM BETWEEN WAVELET,PCA,AND THE PROPOSEDIn accordance with Table (1), the SSIM values for the proposed technique are less than those for both wavelet and PCA. It seems that the structure similarity index in both wavelet and PCA is higher than the proposed fusion technique. At this point, it must be mentioned that the SSIM shows the similarity between the fused and the input images according to the original structure. Since the blurring size used in this paper for all test images is approximately half of the input images, then SSIM shows the similarity according to both blurred and sharp parts of the input images. Therefore, the smaller value of SSIM for the proposed technique does not mean that it is not good. But it exactly means that the degree of similarity for the fused image constructed by the proposed technique is less than the degree of similarity for both the wavelet and PCA. This contribution can be consolidated by the ocular results in Figures (5) to (8). Actually, the visual results obtained by the proposed technique have more sharpening details than other methods (wavelet and PCA).As another performance measure, the PSNR was used to test the goodness of the constructed image. Obviously, the PSNR measure can not be computed unless the original image must be present. This can not be done when using SSIM because the original image may not be always available because the multi-focused images are obtained in different zooming times and angles for the same scene. In fact, there is no perfect image thatcan give the ability to be compared with. Hence, in this paper, the PSNR can not be evaluated for all the test images. Therefore, three test images have their original images present which are Lena, Peppers (V), and Peppers (D), where V stands for Vertical and D for diagonal blurring direction.TABLE2COMPARISON OF PSNR BETWEEN WAVELET,PCA,AND THE PROPOSEDClearly, according to Table (2), the PSNR values forthe proposed technique are higher (in small absolutedifference) than the PSNR values for both the waveletand PCA for the fused images. Actually, this means thatthe differences between the constructed image using theproposed technique and the original image are less thanthose of the wavelet and PCA. This result supports thecontribution that was stated previously which is that theproposed technique produces quite satisfactory resultsaccording to its similar common fusion methods such aswavelet and PCA.(a) (b)(c) (d)(e)Figure 5 (a) Blurred Image (left) (b) Blurred image (right)(c) Wavelet (d) PCA (e) Proposed(a) (b)(c) (d)(e)Figure 6 (a) Blurred Image (left) (b) Blurred image (right) (c)Wavelet (d) PCA (e) Proposed(a) (b)(c) (d)(e)Figure 7 (a) Blurred Image (left) (b) Blurred image (right)(c) Wavelet (d) PCA (e) Proposed(a)(b)(c)(d)(e)Figure 8 (a) Blurred Image (left) (b) Blurred image (right)(c) Wavelet (d) PCA (e) ProposedV.CONCLUSIONThe main and fundamental conclusion that comesfrom the proposed technique is that its simplicity andperformance compared to other complicated methodsthat produce almost the same results by using highlycomplicated mathematical formulas and time consumingtransformations. The only disadvantage of the proposedtechnique is that the fused image is not always perfect.Hence, there must be a condition for the selectioncriterion for the best standard deviation between severalstandard deviations. This may be solved by future work.Further, the number of input images that was used inthis article is two input images only. But this numbercan be generalized for more than two input images andthat is the reason for the name multi-focused imagefusion technique.The results obtained by the highest value of thestandard deviation technique has proven it’s adequacyand efficiency in constructing a fused image from multi-focus images which were taken by different zoomingprocess and times in the camera sensor. Finally, visualeffect and statistical parameters indicate that theperformance of our new method is better its competitorsof the filed.Additionally, one of the best recommendations thatcan be treated as a future work is the implementation ofthe proposed technique on color images than the grayscale images that were used in this article. Actually, thismay facilitate the way for applying the proposedtechnique into video stream.REFERENCES[1]Ben Hamza A., He Y., Krimc H., and Willsky A.,“A multiscale approach to pixel-level imageFusion”, Integrated Computer-Aided Engineering,Vol. 12, pp. 135–146, 2005.[2]Cao, W.; Li, B.C.; Zhang, Y.A. “A remote sensingimage fusion method based on PCA transform andwavelet packet transform”, InternationalConference of Neural Network and Signal Process,vol. 2, pp. 976–981, 2003.[3]Choi M., “A New Intensity-Hue-Saturation FusionApproach to Image Fusion With a TradeoffParameter”, IEEE Transactions On GeoscienceAnd Remote Sensing, Vol. 44, No. 6, June 2006.[4]Choi M., Kim R. Y., Nam M. R., “Fusion of multi-spectral and panchromatic satellite images usingthe curvelet transform”, IEEE Geosci. RemoteSens. Lett., vol. 2, pp. 136–140, 2005.[5]Cunha A. L., Zhou J. P, Do M. N., “The non-subsampled contourlet transform: Theory, designand applications”, IEEE Trans. Image Process, vol.15, p.. 3089–3101, 2006.[6]Das S., Chowdhury M., and Kundu M. K.,“Medical Image Fusion Based On RippletTransform Type-I”, Progress In ElectromagneticsResearch B, Vol. 30, pp. 355-370, 2011.[7]Delleji T., Zribi M., and Ben Hamida A., “Multi-Source Multi-Sensor Image Fusion Based onBootstrap Approach and SEM A lgorithm”, TheOpen Remote Sensing Journal, Vol. 2, pp. 1-11,2009.[8]Do M. N., Vetterli M., “The contourlet transform:An efficient directional multi-resolution imagerepresentation”, IEEE Trans. Image Process, vol.14, pp. 2091–2106, 2005.[9]Flusser J., Sroubek F., and Zitov B., “ImageFusion: Principles, Methods, and Applications”,Lecture notes, Tutorial EUSIPCO 2007.[10]Hill P. R.; Bull D. R., Canagarajah C.N., “Imagefusion using a new framework for complexwavelet transforms”, Int. Conf. Image Process, vol.2, pp. 1338–1341, 2005.[11]Hill P., Canagarajah N., and Bull D., “Imagefusion using complex wavelets”, Proceedings ofthe 13th British Machine Vision Conference,University of Cardiff, pp. 487-497,2002.[12]Ioannidou S.; Karathanassi V., “Investigation ofthe dual-tree complex and shift-invariant discretewavelet transforms on Quickbird image fusion”,IEEE Geosci. Remote Sens. Lett., vol. 4, pp. 166–170, 2007.[13]Klonus S., Ehlers M., “Performance of evaluationmethods in image fusion”, 12th InternationalConference on Information Fusion Seattle, WA,USA, July 6-9, 2009, pp. 1409 – 1416.[14]Li S., Kwok J. T., and Wang Y., “Multifocusimage fusion using artificial neural networks”,Pattern Recognition Letters, Vol. 23, pp. 985–997,2002.[15]Miles B., Ben Ayed I., Law M. W. K., Garvin G.,Fenster A., and Li S., “Spine Image Fusion viaGraph Cuts”, IEEE tansaction on biomedical engineering, Vol. PP, No. 99, 2013.[16]Nencini F., Garzelli A., Baronti S.,”Remotesensing image fusion using the curvelet transform”,Inf. Fusion, vol. 8, pp. 143–156, 2007.[17]Pajares G., De la Cruz J. M., “A wavelet-basedimage fusion tutorial”, Pattern Recognition, Vol.37, pp. 1855 – 1872, 2004.[18]Qiguang M., Baoshu W., Ziaur R., Robert A. S.”Multi-Focus Image Fusion Using Ratio Of BlurredAnd Original Image Intensities”, Proceedings OfSPIE, The International Society For Optical Engineering. Visual Information Processing XIV :(29-30 March 2005, Orlando, Florida, USA.[19]Luo R., Kay M., Data fusion and sensorintegration: state of the art in 1990s. Data fusionin robotics and machine intelligence, M. Abidi andR. Gonzalez, eds, Academic Press, San Diego,1992.[20]Riyahi R., Kleinn C., Fuchs H., “Comparison ofDifferent Image Fusion Techniques For individualTree Crown Identification Using Quickbird Images”, In proceeding of ISPRS Hannover Workshop 2009, 2009.[21]Rockinger O., Fechner T., “Pixel-Level ImageFusion: The Case of Image Sequences”, InProceedings of SPIE, Vol. 3374, No. 1, pp. 378-388, 1998.[22]Song H., Yu S., Song L., “Fusion of multi-spectraland panchromatic satellite images based on contourlet transform and local average gradient”,Opt. Eng., vol. 46, pp. 1–3, 2007.[23]Wan T., Canagarajah N., and Achim A.,“Compressive Image Fusion”, In the Proceedingsof the IEEE International Conference on ImageProcessing (ICIP), pp. 1308-1311, October, 2008,San Diego, California, USA.[24]Wang Z. and Bovik A. C., “A universal imagequality index”, IEEE Signal Processing Letters,Vol. 9, No. 3, pp. 81–84, 2002.[25]Wang Z., Bovik A. C. , “Image QualityAssessment: From Error Visibility to StructuralSimilarity”, IEEE Transactions on Image Processing, Vol. 13, No. 4, pp. 600-612, 2004. [26]Wang Z., Ziou D., Armenakis C., Li D., and Li Q.,“A Comparative Analysis of Image Fusion Methods”, IEEE Transactions On Geoscience AndRemote Sensing, Vol. 43, No. 6, pp. 1391-1402,2005.[27]Yuan J., Shi J., Tai X.-C., Boykov Y., “A Study onConvex Optimization Approaches to Image Fusion”, Lecture Notes in Computer Science, Vol.6667, pp. 122-133, 2012.[28]Zebhi S., Aghabozorgi Sahaf M. R., and SadeghiM. T., “Image Fusion Using PCA In CS Domain”,Signal and Image Processing: An InternationalJournal, Vol.3, No.4, 2012.[29]Zhang L., Zhang L., Mou X. and Zhang D., “FSIM:A Feature SIMilarity Index for Image QualityAssessment”, IEEE Transactions on Image Processing, Vol. 20, No. 8, pp. 2378-2386, 2011. Firas A. Jassim, male, was born in Baghdad, Iraq, in 1974. He received the B.S. and M.S. degrees in Applied Mathematics and Computer Applications from Al-Nahrain University, Baghdad, Iraq, in 1997 and 1999, respectively, and the Ph.D. degree in Computer Information Systems (CIS) from the Arab University for Banking and Financial Sciences, Amman, Jordan, in 2012. In 2012, he joined the faculty of business administration, department of management information systems, Irbid National University, Irbid, Jordan, where he is currently an assistance professor. His current research interests are image compression, image interpolation, image segmentation, image enhancement, image fusion and simulation.。

穿透毛玻璃的可见光成像系统李成勇;应春霞;胡晶晶【摘要】透过复杂介质获取目标物体图像精细信息的能力是光电图像采集处理的一大难点,选用CMOS光电图像传感器,设计了CMOS成像系统以及后端读取和处理电路,透过毛玻璃对目标物体成像,将采集的图像信息传送到计算机中进行处理.该系统按照相机光学成像系统原理制作,采用通用CMOS图像传感器芯片完成电路设计,加之红外激光辅助照明拍摄采集图像,由远及近不同距离分别对同一目标物成像,对成像图像进行迭代图像增强算法优化,可以解决毛玻璃非匀质问题,使光源重建精度大大提高,得到的可见光图像轮廓清晰,与一般CCD成像系统相比,识别率超过95％,远大于一般成像系统,且成像性能良好.【期刊名称】《应用光学》【年(卷),期】2019(040)003【总页数】6页(P416-421)【关键词】可见光;CMOS图像传感器;光电成像;图像处理【作者】李成勇;应春霞;胡晶晶【作者单位】重庆工程学院电子信息学院,重庆400056;重庆工程学院电子信息学院,重庆400056;重庆工程学院电子信息学院,重庆400056【正文语种】中文【中图分类】TN29引言毛玻璃也叫雾面玻璃、防眩玻璃等，是用金刚砂等磨过或以化学方法处理过的一种表面粗糙不平整的半透明玻璃，毛玻璃表面不平整，光线通过毛玻璃被反射后向四面八方射出去，因为毛玻璃表面不是光滑的平面，使光产生了漫反射，折射到视网膜上已经是不完整的像，于是就看不见玻璃背后的图像。

毛玻璃对于不同光波段的成像影响差别比较大，常见的成像光波段有可见光、紫外光、红外光。

紫外光波长较短，吸收大，穿透深度短，常用于透射皮肤等物质成像。

毛玻璃因为有水等少量杂质，紫外光相对于可见光在穿透毛玻璃时有着较高的吸收，因此穿透毛玻璃成像选择透射率高的可见光，毛玻璃对可见光而言表面平整，吸收弱，折射率低。

穿透毛玻璃成像技术，是通过光电图像传感器捕捉视频图像，进行图像信息采集，可以获取人类视觉上看不到的图像信息，让更多的潜在信息被捕捉到。

基于多尺度特征融合网络的多聚焦图像融合技术作者：吕晶晶张荣福来源：《光学仪器》2021年第05期摘要：多聚焦圖像融合技术是为了突破传统相机景深的限制，将焦点不同的多幅图像合成一幅全聚焦图像，以获得更加全面的信息。

以往基于空间域和基于变换域的方法，需要手动进行活动水平的测量和融合规则的设计，较为复杂。

所提出的方法与传统的神经网络相比增加了提取浅层特征信息的部分，提高了分类准确率。

将源图像输入训练好的多尺度特征网络中获得初始焦点图，然后对焦点图进行后处理，最后使用逐像素加权平均规则获得全聚焦融合图像。

实验结果表明，本文方法融合而成的全聚焦图像清晰度高，保有细节丰富且失真度小，主、客观评价结果均优于其他方法。

关键词：多聚焦图像融合;卷积神经网络;多尺度特征中图分类号：TP 183文献标志码：AMulti-focus image fusion technology based on multi-scale feature fusion networkLU Jingjing，ZHANG Rongfu（School of Optoelectronic Information and Computer Engineering，University of Shanghai for Science and Technology，Shanghai 200093，China）Abstract：The multi-focus image fusion technology is to break through the limitation of the traditional camera's depth of field and combine multiple images with different focal points into a full-focus image to obtain more comprehensive information. In the past，methods based on the spatial domain and the transform domain required manual activity level measurement and fusion rule design，which was more complicated. Compared with the traditional neural network，the method proposed in this paper increases the part of extracting shallow feature information and improves the classification accuracy. The source image is input into the trained multi-scale feature network to obtain the initial focus map. Then，the focus map is post-processed. Finally ，the pixel- by-pixel weighted average rule is used to obtain the all-focus fusion image. The experimental results show that the all-focus image fused by the method in this paper has high definition，richdetails，and low distortion，the subjective and objective evaluation results are better than those of other methods for comparison.Keywords：multi-focus image fusion;convolutional neural network;multi-scale features 引言图像融合技术是指将多张图像中的重要信息组合到一张图像中，比单一源图像具有更丰富的细节[1]。

Multifocus image fusion using region segmentation and spatial frequency
xoulongxia 专业：计算机科学与技术
一 研究背景 二 研究动态
三 研究内容
四 文章理解 五想 法
研究背景
• 定义
图像融合（Image Fusion）是通过对源图像间冗余 信息和互补信息进行处理，使得到的融合图像可 靠性增强，能更客观地、更精确地和更全面地对 某一场景进行图像描述，更加适合人眼和机器视 觉感知，更加有利于图像分割、特征提取和目标 识别等更深层次的图像处理任务。
实验方案
▽1、Region segmentation using normalized cuts
（基于归一化分割方法的区域分割） • 被分割的图像是由点集组成的，节点之间的相互联系就构成了加权无 向图G = (V,E)。 • 我们给出 normalized cuts （Ncut）的定义：
Ncut( A, B ) cut( A, B ) cut( A, B ) assoc( A,V ) assoc( B ,V )
• 图像的总的空间频率：
▽3、Multifocus image fusion using regions’ clarity （区域明确的多聚焦图像融合） • 融合的过程 （1）获得检录图像A和B的临时融合图像平均值； （2）使用归一化的分割算法把临时融合图像分割为几个区 域； （3）使用第(2)步的结论划分图像A和B； （4）计算每一个划分的A和B区域的空间频率； （5）比较两个源图像相匹配的区域空间频率，以决定哪一 部分用于构建融合图像，
• 特征级图像融合是中间层次上的融合，它是先提 取来自传感器的原始信息的特征，产生特征矢量, 然后对特征矢量进行融合处理。一般来说，提取 的特征信息应该是像素信息的充分表示量或充分 估计量，然后按照特征信息对多传感器数据进行 综合分析和处理。 特征级图像融合的主要优点有：由于提取传感器原 始信息的特征信息，信息得到了压缩，有利于实 时处理。